AI, Utilities, and Solving Key Business Problems: Part 4 – From Insight to Impact: How Better Data Enables Better AI

The goal of this series has been to take you on a journey. Part 1 showed you that AI isn’t magic. Part 2 established that algorithms are ready, but data quality determines success. Part 3 showed why data needs to be managed as an independent strategic asset, separate from the source systems that store it.

So, now we are here to discuss the big question: How does that translate to real value?

Answer: Through connection.

The Pole, Evolved

We used a pole as an example in Part 3, so let’s take that further. Today, utilities manage poles using institutional knowledge and hard work. Enterprise Asset Management knows the asset’s provenance and condition, GIS has the details about materials and location, the Customer Information System knows who pays the bills, and repair records might be in many places (or on paper). There is probably someone with a lot of experience who has the complete picture, but it’s not in any one system.

Now imagine if all that information was connected. Not moved to a single database or dumped into a data lake but connected in a way that preserves context and meaning.

A pole isn’t just a GIS record; it’s a node in a network. It connects to lines, transformers, customers, work history, inspection records, environmental conditions, and the thousands of decisions that keep the grid running. When that pole’s data exists as a true reflection of those relationships, something changes.

A good data architecture connects these systems, creating trusted connections across information that was previously fragmented. When you understand, trust, and govern your data you establish the foundation required for AI to work as designed.

Let’s take our pole example from our previous discussion one step further. Once that information is connected and governed, it can be represented in a graph, where the pole exists not as a single record in a single system, but as part of a network of relationships linking assets, customers, work history, inspections, environmental conditions, and operating events. This isn’t magic. It’s the logical evolution of treating data as a strategic asset.

With those relationships explicitly connected, AI can evaluate questions that were previously difficult or impossible to answer. “Which poles like this one failed during previous storms?” “Which poles haven’t been inspected since 2018 and have unknown attachment loads?” “Show me poles with similar characteristics located in high-wind zones.” The question is no longer just “What is this pole?” but “What does this pole mean in the context of everything connected to it?”

That’s where the combination of trusted data, connected relationships, and the right AI technology begins to create value. The AI isn’t generating insight from thin air. It’s doing what we discussed in Part 1 – pattern recognition – but adding context to reveal patterns that were previously hidden.

The Workflow That Matters

Once you have that complete picture, the AI model you deploy becomes dramatically more valuable. It trains on reliable, contextual data. It can identify more complex patterns and surface deeper correlations. It can predict which poles might be most at risk, not in isolation, but in the context of their role in the grid.

When this happens, the utility has the potential to avoid reacting to failures and instead start preventing them. Work orders get prioritized based on actual risk, not guesswork. Crews can potentially be dispatched to the poles most likely to fail during the next storm, before the storm hits. When something does fail, there is information that helps determine a complex root cause, which can help you predict and prevent similar failures elsewhere.

It isn’t AI doing something magical, but doing what it was designed to do, pattern recognition at scale. The right foundation brings together a business problem, curated enterprise level data, and the right tools.

Why It Matters (And Why It’s Repeatable)

The pole isn’t unique. The same pattern works for vegetation management. The data is connected: imagery, GIS location, inspection history, work records, and environmental data. AI identifies potential high-risk trees so they can be evaluated before they hit lines.

It works for demand forecasting. The data is connected: customer consumption data, weather patterns, solar generation, EV charging trends, and historical events. AI predicts demand more accurately, reducing strain on the grid.

It works for crew dispatch, asset replacement planning, outage prediction, and every other decision that matters to a utility. This is playing out in the real-world right now, with organizations like Evergy leveraging this framework for monitoring and diagnostics^[1] and Volue, a Norwegian company serving utilities, using graphs to improve the grid it managed^[2].

The pattern is repeatable. Understand your decision, assemble the information required to make it well, connect that information in a way that preserves context, and deploy AI with confidence. It’s the framework from the previous three articles, now in action.

The foundation that has been laid to create trusted data and connected data starts to pay off as you scale. Treating data as a strategic asset isn’t abstract. It’s the practical foundation that makes AI actually work.

What Success Actually Looks Like

You don’t need the most sophisticated model. You don’t need to rebuild your entire data infrastructure overnight. Start with one key business problem that makes sense for your organization. It could be poles, or vegetation, or forecasting- whatever fits your organization. Then follow the pattern. Understand the problem, curate the data, structure the governance, and when that is established, you deploy an AI solution that’s right-sized for the problem.

The initial model doesn’t have to be perfect. It just needs to move the needle from what you’re doing now, and it will improve as you feed results back into it.

While it’s not always the most visible or flashy, the value is in the prep work we’ve been talking about. The disciplined, methodical work of connecting data that matters to the problems you’re trying to solve.

The Path Forward

No one is asking, “should we invest in AI?” We need to know “how can we get value?” and “are we ready?” And by now, you know the answer depends on whether you’re treating data as a strategic asset, whether you understand the connections your decisions require, and whether you’re set up to do the foundational work.

The technology is out there bringing value when the methods are right, and utilities are proving it every day. The best way to move forward is to focus on the foundations while keeping the long-term vision in mind.

At UDC, that’s exactly where we focus. We help utilities understand their critical decisions, assemble the data those decisions require, create the connections that preserve meaning, and deploy AI solutions that actually work.

The algorithms are ready, and your data can be too.

Missed parts 1, 2, or 3 of TJ’s AI series? Read them here:

Part 1: AI Isn’t a Magic Wand

Part 2: AI by Any Other Name

Part 3: Why We Need to Think About Data Differently